Related Experiment Video
Updated: Mar 18, 2026
08:56
Automatic Image Processing to Determine the Community Size Structure of Riverine Macroinvertebrates
Published on: January 13, 2023
2.9K
Automated otolith image classification with multiple views: an evaluation on Sciaenidae.
J Y Wong1, C Chu1, V C Chong1,2
1Institute of Biological Sciences, University of Malaya, 50603, Kuala Lumpur, Malaysia.
Journal of Fish Biology
|July 2, 2016
Summary
Combining multiple 2D views of fish otoliths significantly improves classification accuracy for Sciaenidae species. This approach enhances automated fish identification systems, making them more robust and efficient.
Area of Science:
- Ichthyology
- Morphometrics
- Computational Biology
Background:
- Fish classification is crucial for fisheries management and biodiversity assessment.
- Traditional methods often rely on external morphology, which can be variable.
- Sagittal otoliths offer detailed shape information for species identification.
Purpose of the Study:
- To evaluate the effectiveness of combining multiple 2D views of sagittal otoliths for fish classification.
- To compare different shape description methods for otolith analysis.
- To develop a generic content-based image retrieval (CBIR) system for otolith image searching.
Main Methods:
- Combined multiple 2D views (proximal, anterior, ventral) of sagittal otoliths.
- Applied shape description methods: shape indices, Procrustes analysis, and elliptical Fourier analysis.
- Developed a CBIR system using Procrustes distance for otolith image retrieval.
Main Results:
- Combined 2D views significantly improved classification accuracy compared to single views for nine Sciaenidae species.
- Procrustes analysis and elliptical Fourier analysis outperformed shape indices with single views.
- All shape description methods performed equally well with combined views.
- The CBIR system successfully retrieved otolith images without requiring specific orientation information.
Conclusions:
- Combining multiple 2D otolith views is a superior method for fish classification.
- Automated CBIR systems using otolith shape are effective for fish identification.
- This approach facilitates efficient and accurate fish species identification in ichthyological research.
